Search Linux Wireless

Re: RSS configuration in iwlwifi

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 2016-04-20 at 19:22 +0300, Emmanuel Grumbach wrote:
> On Wed, 2016-04-20 at 16:47 +0100, Ben Hutchings wrote:
> > On Wed, 2016-04-20 at 15:30 +0000, Grumbach, Emmanuel wrote:
> > > Hi Ben,
> > > 
> > > 
> > > Thanks for looking at our code.
> > > 
> > > 
> > > On Wed, 2016-04-20 at 16:08 +0100, Ben Hutchings wrote:
> > > > 
> > > > I'm not sure if you were aware, but there is a standard API for
> > > > configuring RSS in network drivers, part of ethtool_ops.  I
> > > > think
> > > > iwlwifi should implement that rather than a driver-specific
> > > > debugfs
> > > > interface.
> > > > 
> > > You are right, this is why Sara made this commit:
> > > 
> > > commit 854d773e4ab5869200004af4ca5d851730849903
> > > Author: Sara Sharon <sara.sharon@xxxxxxxxx>
> > > Date:   Tue Mar 22 15:55:58 2016 +0200
> > > 
> > >     iwlwifi: mvm: improve RSS configuration
> > >     
> > >     Improve current RSS configuration:
> > >      * Use netdev_rss_key instead of keeping a local copy.
> > >      * Configure also UDP hashing to have UDP traffic spread
> > > across
> > > queues.
> > >      Do not direct RSS traffic to our fallback queue.
> > > 
> > >     Signed-off-by: Sara Sharon <sara.sharon@xxxxxxxxx>
> > >     Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@xxxxxxxxx
> > > >
> > 
> > That doesn't really address what I said.  Yes, it's using the
> > common
> > RSS key, but it's not implementing the ethtool operations to get
> > and
> > set the indirection table and the types of flow hashing that are
> > enabled.
> > 
> 
> Hm.. I think that setting the indirection table is a problem in our
> case because the PN check is done in the driver. The PN check is the
> way WiFi addresses the replay attack, and since the PN check relies
> on
> per-cpu variables, we cannot *safely* allow users to modify the
> indirection table while traffic is flowing.
> 

Oh - and yes, we do have a debugfs hook for that. It was meant for the
very early testing phases. I guess that we should remove it at some
point, unless we decide that whoever wants to play with the indirection
table while inbound traffic is flowing puts himself at risk.

> > Ben.
> > ��.n��������+%������w��{.n�����{���zW����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f




[Index of Archives]     [Linux Host AP]     [ATH6KL]     [Linux Wireless Personal Area Network]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [Linux Kernel]     [IDE]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite Hiking]     [MIPS Linux]     [ARM Linux]     [Linux RAID]

  Powered by Linux